pects for the developing countries. Additional technical assumptions may also be introduced, depending on the study, but they usually lack empirical foundation and are not central to the conclusion this approach yields. Though it may be untenable to treat population growth as exogenous in the analysis of economic development, the incorporation of population growth into economic growth models remains primitive and controversial because we lack empirical facts on what are the critical interactions, what are the magnitudes of these relationships, and what nontraditional sources of economic growth account for the growth in labor and capital productivity in recent decades.* Without the empirical or theoretical foundation to answer any of these questions, the air of precision attained by such macro-analytic techniques is not particularly helpful today, and may be misleading.t Moreover, one simply cannot consider per capita income as an adequate indicator of personal welfare for evaluating the effects of demographic trends. As emphasized here, children are much more to parents than a source of future earnings; they also yield intangible and important rewards to parents that elude national income accounts.? Nonpecuniary returns can be neglected without harm when one chooses between investments in fertilizer or cement factories, but when the choice is between fewer children and a factory, neglect of nonpecuniary returns biases the selection toward programs and policies that seek to reduce births. Policies that seek to make direct changes in the reproductive preferences of parents should therefore be based on evidence *Leibenstein (in this volume) questions the usefulness of this approach unless it is accompanied by a more completely specified understanding of the economic growth process, including nontraditional inputs. I Myrdal disparages the value of this second approach as currently pursued. This kind of analysis (macro-growth models), in terms of capital investment and output, is typical of the modern approach to the economic problems of South Asian countries. In our opinion it is too mechanistic and schematic. It gives the appearance of knowledge where none exist, and an illusory precision to this pretended knowledge. . .. Increased attention has recently been paid to the economic effects of the change in age distribution resulting from reduced fertility rates, and in consequence more complicated models have been constructed [reference to Coale and Hoover (45)]. . . . But despite their [authors of demographic-economic macro models] disclaimers, they create a false air of precision and a false confidence in their model. A critical examination is therefore justified . . . [and in conclusion] We would not exclude the possibility of constructing models more adequate for this purpose. But such models would have to contain many more parameters and account for many more interrelationships [than Coale and Hoover have]. They would have to be very much more complex in order to be logically consistent and adequate to reality. With the present dearth of empirical data, indulging in this type of preparatory macro-analysis docs not seem to be a rewarding endeavor. (1, Vol. Ill, Appendix 7, pp. 2067-2075.) See also the chapter by Leibcnstein in this volume. tMore narrowly, national income per capita is a poor index for evaluating population policies because consumption requirements are smaller for children than adults. A given level of per capita income clearly means a more satisfactory living standard in an economy where population is growing rapidly and a large fraction of the population are children than in an economy with a low birth rate and a small fraction of children.